Architecture for Human Supervised Autonomously Controlled Off-road Equipment

摘要

Human supervised autonomously controlled systems, also known as telerobots, arecapable of focused perception and intelligent action. In a telerobot, a human operator generatestasks, and a computer autonomously closes some of the controlled loops. However, theacceptance of telerobotic systems is now restricted to DDD (dirty, dangerous and distant)missions, while fully autonomous systems proved capabilities only in elementary tasks. In orderto adapt telerobotic systems to perform complex tasks at a higher level of autonomy, a bridgeshould be built between two distinct, but complementary communities: one dealing withteleoperated technologies and the other with autonomous moving vehicles. An evolutionarysolution may be achieved by adapting the well-accepted NASREM/ Real-time Control System(RCS) telerobotic architecture to the newly emerging paradigm of Behavior Based Systems(BBS) architecture developed for multiple robots and autonomously moving vehicles.Nevertheless, RCS has deliberative capabilities it also enables rapid response to contingentevents, based on preliminarily prepared alternative plans or human operator intervention. Thetask decomposition in RCS is functional and time scale related. On the other hand, somecomplex tasks may be efficiently accomplished by elementary actions, achieved throughbehavioral task decomposition. A suggested combined architecture will be discussed, in which,behavior based software agents acting across some layers of RCS, drive the system towardlocal goals, while the mission is achieved through control arbitration agents.